qwen3-8b-kto-avg-beta0p01-s42/README.md

library_name, base_model, tags

library_name	base_model	tags
transformers	Qwen/Qwen3-8B	ronpo mnpo qwen3 preference-optimization

qwen3-8b-kto-avg-beta0p01-s42

Research checkpoint for the RONPO AAAI revision experiments.

• Method: KTO-avg beta=0.01
• Base model: Qwen/Qwen3-8B with non-thinking generation protocol
• Seed: 42
• Local source at upload time: /NHNHOME/WORKSPACE/26msit001_A/BASE/aipr_lab_sjkim_eval/revision_qwen3_8b/full_iter1/train/kto_avg_beta0p01_s42_nhn-b200-q3-seq-07101430-kto/checkpoint-292
• Uploaded at UTC: 2026-07-11T01:15:32Z
• Run status metadata: not found

Qwen3-8B KTO baseline on the averaged three-reward oracle.

Intended use: reproducibility and evaluation for the RONPO paper. This checkpoint is not intended as a production assistant.